What type of study is this?

November 30, 2025

Comparison of clinical outcomes of sepsis patients in two county emergency departments using systemic inflammatory response syndrome versus Epic’s proprietary severe sepsis alert

Puntos clave

Sepsis alerting via either method shows poor performance with sensitivity around 15%.
In-hospital death rates were 11% under SIRS+ alerts and 8% with ESPMv1, indicating no statistical significance.
Analysis of 881,201 ED patients revealed no significant differences in antibiotic initiation or ICU admissions.
Both alerting systems adequately identified patients but may need enhancement for real-time emergency care.

Resumen

Abstract Objectives To compare the clinical outcomes of sepsis patients when an augmented systemic inflammatory response syndrome (SIRS+) and the Epic sepsis predictive model version 1 (ESPMv1) alert were active in the emergency department at two county hospitals. Materials and Methods This retrospective study from January 2018 to January 2024, evaluated the clinical outcomes of 881 201 emergency department patients of which 29 852 patients were septic. From January 2018 to June 2022 sepsis notices were presented to physicians based on a SIRS plus organ dysfunction criteria and from December 2022 to January 2024 using the ESPMv1 alert. Sepsis was defined according to the Sepsis-3 definition with the onset of sepsis defined as two or more points on the Sequential Organ Function Assessment (SOFA) score in patients where physicians ordered at least one blood culture and antibiotic. SIRS+ alerting occurred when 2 of 4 criteria was reached plus one organ dysfunction measurement. The ESPMv1 alerting occurred at the Epic recommended threshold of six. We evaluated the times to blood cultures, antibiotics and ICU admission requests, and in-hospital death rates. Results SIRS+ alerts had a sensitivity of 14.25%, specificity 96.1%, positive predicative value (PPV) of 7.8% and negative predative value (NPV) of 98%. The ESPMv1 had a sensitivity of 15.6%, specificity 95.4%, positive predictive value of 8.1%, and negative predictive value of 98% for diagnosing sepsis. No statistical differences in time to antibiotics (5.1 vs 5.9 h), time to blood culture draws (3.6 vs 3.5 h) or time to ICU admission (10.4 vs 9.6 h) were observed. We did observe a difference in hospital death rates between the two time periods (11% vs 8%) but no statistical difference when adjusting for unvaccinated covid-19 (OR 0.95 0.87-1.03). Discussion and Conclusion No statistically significant clinical differences or performance metrics were observed between SIRS+ based alerting and ESPMv1 alerts in an undifferentiated emergency department population. Both alerting systems had similarly poor diagnostic characteristics.

Me gusta

Guardar